The Journal of Neuroscience, October 1989, 9(10): 34833481

Morphology and Electrophysiological Properties of lmmunocytochemically Identified Rat Dopamine Neurons Recorded in wifro

Anthony A. Grace and Shao-Pii Onn Departments of Behavioral Neuroscience and Psychiatry, Center for Neuroscience, University of Pittsburgh, Pittsburgh, Peksylvania 15260

In vitro intracellular recordings were made from neurons in hyperpolarizations, (3) high threshold dendritic calcium the rat midbrain slice. Two neuronal types could be distin- spikes which gave rise to the spike afterhyperpolariration, guished in dopamine-containing (DA) midbrain regions based and (4) a high threshold initial segment sodium spike. These on electrophysiological criteria. One neuron type exhibited depolarizations were modulated by several processes, in- short duration action potentials (< 1.5 msec), could fire at cluding a 4-aminopyridine-insensitive delayed repolariza- high frequencies (> 10 Hz), and exhibited either phasic or tion, an instantaneous and time-dependent anomalous rec- burst firing patterns. This neuron did not exhibit tyrosine tifier, and an afterhyperpolarization. Although low threshold hydroxylase immunoreactivity. A second neuronal type ex- depolarizations and rebound action potentials could be trig- hibited a unique set of electrophysiological properties, which gered by the membrane repolarization following small mem- included (1) a spontaneous pacemaker-like depolarizing po- brane hyperpolarizations, comparatively larger hyperpolar- tential, (2) a highly regular firing pattern, (3) long duration irations attenuated this rebound activation, thereby (>2 msec) action potentials, and (4) a high (i.e., depolarized) suppressing anodal break excitation. spike threshold. This neuron was consistently double la- These experiments provide a basis for the identification beled using intracellular staining and immunocytochemical of DA neurons in the in vitro brain slice preparation by their localization of the catecholamine-specific enzyme tyrosine characteristic electrophysiological properties. By examining hydroxylase, and thus represented the DA neuronal type. the morphology and regulation of activity in this neurochem- Midbrain DA neurons stained with Lucifer yellow could be ically defined cell class, information pertaining to the func- separated into 3 classes based on their location and mor- tioning of DA neurons will be garnered, which may eventually phology: (1) fusiform neurons with laterally projecting den- lead to alternate avenues of therapeutic intervention in DA- drites in the dorsal substantia nigra zona compacta region, related psychiatric disorders. (2) multipolar cells with laterally and ventrally projecting den- drites in the ventral substantia nigra zona compacta, and (3) Dysfunctions of the midbrain dopaminergic (DA) systems have neurons with fusiform and multipolar somata and radially been implicated in the etiology of several clinical disorders, such projecting dendrites in the ventral tegmental area. The den- as Parkinson’s disease (Homykiewicz, 1963, 1966) and schizo- drites also exhibited spine-like protrusions and ended with phrenia (Snyder, 1973; Matthysse, 1973). These findings have specialized forked processes. led to extensive research into the function of this small group Spontaneously firing DA cells recorded in vitro had a num- of brain stem neurons using several experimental preparations. ber of distinguishing electrophysiological characteristics in Among the approaches used to investigate this system, in vivo common with those of DA neurons recorded in vivo, such as electrophysiological recordings from identified DA-containing the presence of a slow depolarizing potential driving spike neurons have played a major role in determining the mode of activity and a characteristic depolarized spike threshold (ap- action of a variety of psychotogenic and therapeutic drugs, such proximately -36 mV). However, in contrast to that found in as amphetamine (Groves et al., 1975; Bunney and Aghajanian, viva, the DA cells characterized here exhibited substantially 1976) and the neuroleptics (Bunney and Grace, 1978; Chiodo higher input resistances and fired spontaneously in a very and Bunney, 1983; White and Wang, 1983; Grace and Bunney, regular pacemaker pattern. Burst firing was not observed. 1986). However, in order to derive valid neuropharmacological Spike activity was apparently dependent on 4 depolarizing information about the mechanisms of drug action through the events: (1) a voltage-dependent TTX-sensitive slow depo- recording of neuronal activity, it is first necessary to establish larization, (2) a cobalt-sensitive low threshold depolarization that the cells recorded contain the neurotransmitter under study. that was activated during the rebound from brief membrane By combining electrophysiological characterizations of putative midbrain DA neurons with histochemical confirmation of their neurochemical identity (Bunney et al., 1973; Grace and Bunney, Received Jan. 13, 1989; revised Mar. 8, 1989; accepted Apr. 6, 1989. 1980, 1983a), a set of physiological criteria was established for We wish to thank Mr. Jeffrey Hollerman and Ms. Michele Pucak for suggestions in the preparation of this manuscript. This work was supported by USPHS identifying DA neurons recorded in situ in the anesthetized rat. MH42217, MH30915, NS19608, and the Scottish Rite Schizophrenia Research In this way, the in vivo studies of DA neuron activity have Program, N.M.J., U.S.A. S.P.O. is supported by a Tourette’s Foundation fellow- ship. A.A.G. is a Sloan fellow. provided information about the range of responses and dis- Correspondence should be addressed to Anthony A. Grace at the above address. charge patterns that DA neurons are capable of expressing in Copyright 0 1989 Society for Neuroscience 0270-6474/89/103463-19$02.00/O the intact organism and how this activity could be modulated 3464 Grace and Onn - Identification and Characterization of DA Neurons in vitro by drugs administered to the whole organism. However, in viva Laboratory Animals published by the USPHS. The specific protocols recordings are somewhat limited in their ability to identify the used were approved by the University of Pittsburgh Animal Care Com- mittee. Rats weighing between 225 and 350 gm were deeply anesthetized membrane processes underlying neuronal activity or in distin- with sodium pentobarbital (Nembutal; 50 mg/kg i.p.) before decapita- guishing between pharmacological responses mediated directly tion. After removal of the surrounding cranium, the brain was rapidly and those acting through afferent systems. Thus, we initiated a excised, and a 4 mm thick section was cut perpendicular to the brain series of studies to investigate the membrane properties that axis using an Activational Systems Rat Brain Matrix (RBM-4000C) to underlie the activity of midbrain DA neurons using the in vitro guide the cut. A block from the midbrain region (2200-4700 pm anterior to the interaural line; Paxinos and Watson, 1982) was then sectioned brain slice preparation, which is more amenable to this level of in ice-cold Ringer’s solution (124 mM NaCl, 5 mM KCl, 1.2 mM KH,PO,, analysis. 2.4 mM CaCl,, 1.3 mM MgSO,, 26 mM NaHCO,, 10 mM glucose, and One problem that arises in using the in vitro preparation for saturated with 95%:5% O,:CO,) to 400 pm thickness using a Vibratome investigating DA neuron electrophysiology is again confirming (Oxford). The brain sections were placed in continuously oxygenated Ringer’s solution at room temperature for l-2 hr before recording. The that the recordings were made from identified DA neurons. slices were then transferred to a submersion-type chamber maintained Unfortunately, many of the identification criteria advanced in at 37°C and superfused with oxygenated Ringer’s adjusted to a flow rate the in vivo experiments could not be applied to the in vitro of 2-3 ml/min, as previously described (Llinls and Sugimori, 1980a, preparation, such as the ability to antidromically activate neu- b; Grace and Llinas, 1985). rons from their termination sites (Deniau et al., 1978; Guyenet Signals captured by the electrodes were amplified by an adjacent head stage amplifier connected to a preamplifier (NeuroData IR-283). Current and Aghajanian, 1978; Thierry et al., 1979; Preston et al., 198 1; was injected across a bridge circuit, with electrode voltage and current Grace and Bunney, 1980, 1983a). Other electrophysiological injection amplitudes monitored on an oscilloscope (Hitachi V- 105OP). criteria used to identify DA neurons in vivo, such as firing pat- These output signals were also digitized at 44 kHz per channel using a tern, appear to be sufficiently altered in vitro (Brodie and Dun- NeuroData 4-channel Neurocorder (Neurodata DR-484) and stored on VHS videotaues (VHS T-120) for subseauent off-line analvsis. Strike widdie, 1987; Grace, 1987; Shepard and Bunney, 1988) to ren- potentials were analyzed using a CED1401 Intelligent Laboratory In- der this classification criterion inadequate for this preparation. terface (Cambridge Electronic Design, Cambridge, UK) outfitted with Identification based on anatomical location is complicated by 8 megabytes of Massram. This unit was interfaced with a Compaq 386/ many factors, including (1) the observation that DA neurons in 20 microcomputer equipped with a 130 MB high-speed fixed disk drive the zona compacta lie within a thin sheet of cells and are in- and an Amdek 1280 high-resolution monochrome monitor for display, measurement, and analysis of the digitized data. Hard copy displays of termixed with non-DA neurons (Guyenet and Crane, 198 1) and long spike trains were produced by directing the output of data captured (2) a study reporting a significant probability of impaling and by the CED1401 Massram to a Gould RS3400 3-channel chart recorder. staining non-DA neurons in the substantia nigra zona compacta In order to assure accurate representation of spike peaks and fast tran- during in vitro recordings (Bargas et al., 1988). Therefore, the sients, the digitized data stored in the CEDl40 1 Massram were con- verted into analog form by running the CED1401 internal digital-to- first step in this investigation was to identify, in the in vitro slice analog converters at 1/ 100 of the rate used for data capture, thus assuring preparation, which physiologically defined neuronal type in the that the output transients would remain within the limited dynamic substantia nigra region contained DA as a neurotransmitter. range of the chart recorder pens. Details of shorter duration spike trains The next stage of analysis involved the development of a or responses to individual current pulses were obtained by directing model of DA neuron action potential generation in terms of the data through the Compaq computer for output onto a Hewlett-Packard 7475A plotter using drafting pens and polyester film to generate accurate cell’s spike generating sites and their role in promoting neuronal plots of the stored data. discharge. This would be important for understanding such cel- Electrodes were pulled from 1 mm diameter Omegadot (WPI, New lular properties as (1) the manner by which excitation is spread Haven, CT) borosilicate glass tubing using a Flaming/Brown P-8O/PC throughout the neuron, (2) the sites at which synaptic input may electrode puller. The glass pipettes were then filled with one of the following solutions: (1) 3.0 M potassium acetate for intracellular re- be positioned to exert a significant influence on cell activity, (3) cording (impedance, 45-75 MB) or (2) 10% Lucifer yellow (Sigma; Stew- how modulation of active membrane sites within a neuron may art, 1978) in distilled water (impedance, 80-150 MB) for combined influence its firing rate or pattern, and (4) whether zones of recording and staining. Current was injected across a bridge circuit built current influx are related to release of neurotransmitters from into the preamplifier. Only neurons evidencing stable penetrations were used for electrophysiological analyses and dye injection, with stability dendritic sites. This type of information may be highly relevant defined as resting potentials greater than -50 mV, action potential for discerning the functional implications of DA cell electro- amplitudes greater than 65 mV, and temporally uniform input resis- physiological activity, given that these cells are known to contain tances typically greater than 80 MB. Putative DA neurons were tenta- and release DA from their dendritic as well as their axon ter- tively identified during intracellular recording by their location, the minal regions (Geffen et al., 1976; Korfet al., 1976), and there presence of a slow pacemaker-like depolarization (slow depolarization) preceding spontaneously occurring action potentials, and the long du- is evidence suggesting that dendritic DA release is likely to be ration (>2 msec) action potentials (Grace and Bunney, 1983a). Intra- modulated by factors different from those regulating the release cellular staining was performed using electrodes filled with Lucifer yel- of DA from nerve terminals (ChCramy et al., 198 1, 1985; Romo low dissolved in distilled water, and tyrosine hydroxylase et al., 1986). Thus, details of DA neuron spike generation may immunoreactivity was localized using an indirect immunofluorescence be relevant for understanding how DA cell activity states and technique. Cells were stained intracellularly using electrodes filled with 10% Lucifer yellow (Sigma), with the dye ejected by applying l-2 nA firing patterns are regulated and, furthermore, can provide in- continuous hyperpolarizing current to the electrode. This constant cur- sights into the relationship between these spike-generating zones rent was interrupted by 50-msec 2 nA depolarizing current pulses de- and the compartmentalization of function within this neuronal livered at 5-6 Hz to prevent clogging of the electrode tip (Grace and type. Part of these data has been presented at symposia (Grace, Llinas, 1985). Cells were injected with the dye for 5-20 min to obtain details of distal processes for use in morphological studies and for l-4 1987, 1988; Grace and Onn, 1988a, b). min in the double-labeling studies to circumvent bleedthrough of Lucifer yellow fluorescence at longer wavelengths. Slices used in double-labeling Materials and Methods studies were then placed in fixative (4% paraformaldehyde in 0.1 M Sprague-Dawley albino male rats obtained from Zivic-Miller Labora- phosphate buffer, pH, 7.4, containing 10% sucrose) overnight at 4°C. tories were used in all experiments and were handled in accordance Serial sections 20 pm in thickness were cut on a cryostat (Reichert) and with the procedures outlined in the Guide for the Care and Use of collected onto albumin-coated slides. The Lucifer yellow-injected cells The Journal of Neuroscience, October 1989, 9(10) 3485 werelocalized in the sectionsusing a Leitz Orthoplan II epifluorescence cated in the substantia nigra zona compacta, the nearby reticular microscopeequipped with a Leitz D filter cube (excitation: band pass formation, and the zona reticulata. These cells had numerous 355-425 nm; dichromatic mirror: RKP 455 nm; suppression:low pass 460 nm). Typically, the entire soma of the stained neuron was located dendritic processes arising directly from the soma, including in a single section,with distal processescontained in adjacentsections. dendrites extending into regions dorsal to the substantia nigra. Slicesused for morphologicalinvestigations were fixed overnight in 10% In 3 cases, neurons of this class that had been stained by intra- phosphate-bufferedformalin (pH 7.4), washed in buffered saline, and cellular injection of Lucifer yellow were examined for the pres- clearedbv ulacine;them in 100%dimethvlsulfoxide (DMSO) for 20 min ence of tyrosine hydroxylase-like immunoreactivity using a rho- (Grace and Llin& 1985). Slices were ihen transferred to ‘microscope slidesand mounted in DMSO for viewing with the fluorescencemicro- damine-labeled antibody. None of these stained neurons were scope. found to exhibit double labeling for tyrosine hydroxylase im- Tyrosine hydroxylase-like immunoreactivity was localized using im- munoreactivity. Furthermore, the morphological, histochemi- munocytochemicaldetection (Stembergeret al., 1970) on sections of cal, and electrophysiological characteristics found for this neu- brain slices containing the Lucifer yellow-labeled cells and their pro- cesses.The slides were incubated with tyrosine hydroxylase antisera ron type were not consistent with those of histochemically (EugeneTech; 1000x dilution with 0.1~ phosphatebuffer containing identified DA neurons described in anatomical studies (Schwyn 5%normal goat serumand 0.3%Triton X- 100)for 2 d in a moisturized and Fox, 1974; Hiikfelt et al., 1976; Lindvall and Bjorklund, chamber maintained at 4°C followed by overnight incubation with 1974, 1978; Juraska et al., 1977) or in vivo intracellular recording rhodamine-conjugatedgoat antirabbit IgG (Sigma; 60 x dilution with and staining of physiologically identified DA cells (Grace and samebuffer), also at 4°C. Sectionswere washed repeatedly,then air- dried and mounted in glycerol. Double-labeled cells were first photo- Bunney, 1980, 1983a, b; Grace, 1987, 1988; Tepperet al., 1987). graphed using a Leitz N2 filter (excitation: band pass 530-560 nm; DA neuron identification. A physiologically distinct neuronal dichromatic mirror: RKP580 nm; suppression:low pass580 nm), which type was also recorded in the zona compacta region of the sub- passesfluorescence from the rhodamine label and attenuatesthat from stantia nigra and in the ventral tegmental area. These neurons the Lucifer yellow. The cells were photographedagain after changingto the Leitz D filter, which will preferentially passLucifer yellow fluores- exhibited action potentials similar to those reported for iden- cenceand block fluorescencearising from the rhodamine label. tified DA neurons recorded in vivo; i.e., action potentials (1) The pharmacologyof the membranepotential changesobserved was were long in duration (> 2 msec), (2) had comparatively depo- determinedby the-addition of specific-ion channel blockers [l -2 WM larized thresholds for activation (i.e., -30 to -45 mV), (3) were TTX to block sodium channels:24 mM tetraethvlammonium (TEA) triggered by a voltage-dependent slow depolarization, and (4) to block some potassium channels; l-3 mM cobalt to block calcium influx; and 4-aminopyridine (4-AP): 5 mM and 20 mM to block pref- were followed by a prominent afterhyperpolarization (Fig. 1C). erentially the transient type potassium channels]. These compounds In every case where neurons displaying these electrophysiolog- were addedto the Ringer’s at a constant rate set by a microprocessor- ical properties were stained by intracellular injection of the dye controlled peristaltic pump (Haake-Biichler model MCP2500), which Lucifer yellow, the neurons also demonstrated double labeling added the solutions directly to the mixing compartment of the slice chamber.This permitted rapid and precisecontrol of the composition for tyrosine hydroxylase-like immunoreactivity (n = 7), both in of the medium. For studiesinvolving the administration of cobalt, Tris the substantia nigra (Fig. 2A) and in the ventral tegmental area buffer was substitutedfor the bicarbonatebuffer to avoid precipitation (Fig. 2B). Of course, tyrosine hydroxylase is not specific for DA of the cation, and the solution was oxygenatedusing 99.5% 0,:0.5% neurons, since this enzyme is also located in other catechol- co,. amine-synthesizing neurons, i.e., those containing norepineph- All values given in text are mean f SD. rine or epinephrine. However, studies have failed to localize the specific noradrenergic enzyme dopamine-P-hydroxylase (Fuxe Results et al., 197 1; Swanson and Hartman, 1975) or the adrenergic Cell identijication enzyme phenylethanolamine N-methyltransferase (Kitahama Intracellular recordings were made from neurons in the zona et al., 1988) to neuronal somata in this brain region. Further- compacta region of the substantia nigra and in the ventral teg- more, rhodamine-labeled cells were only observed in midbrain mental area. Neurons recorded within these midbrain regions regions known to contain DA cells (i.e., the zona compacta of could be readily classified into two broad subtypes based on the substantia nigra and the ventral tegmental area). Thus, these their morphological and electrophysiological characteristics. Only experiments confirm that this class of physiologically defined one of these neuron types was labeled as containing tyrosine neurons was of the DA-containing type. hydroxylase-like immunoreactivity. One problem often encountered with double-labeling studies Non-DA neurons. One neuronal type recorded within these is distinguishing the fluorescence of the injected dye from the regions rarely exhibited spontaneous spike activity and, on de- fluorescence of the immunocytochemical stain. In these exper- polarization, fired large amplitude, fast action potentials (60- iments, care was taken to ensure that comparatively small 75 mV, duration: 1.3 + 0.3 msec, mean + SD). The action amounts of Lucifer yellow were injected into the neuron, thereby potential threshold was typical ofmammalian neurons (i.e., - 53 limiting the observable emission spectrum of this dye. Thus, in +- 4 mV) and did not vary appreciably with changes in mem- double-labeling studies, Lucifer yellow was injected for periods brane potential. The input resistance of the neurons averaged of only l-4 min to facilitate its discrimination from the im- 135 + 50 MB (n = 12). Depolarization of the membrane from munocytochemical fluorescent label. Under these circum- resting potentials elicited repetitively firing spikes (Fig. L4). stances, only neurons displaying the characteristic DA cell phys- However, depolarization of the membrane from more hyper- iological pattern demonstrated double labeling for Lucifer yellow polarized potentials (i.e., negative to - 75 mV) elicited a low and tyrosine hydroxylase immunoreactivity. Bleedthrough of threshold, TTX-resistant depolarization with an accompanying the Lucifer yellow dye across the rhodamine filter was not a burst of action potentials. The interval between consecutive confounding variable in these experiments because (1) non-DA spikes occurring within a burst was very brief (i.e., 3-5 msec; neurons in the zona reticulata or reticular formation injected Fig. lB). Intracellular injection of Lucifer yellow into neurons with an equivalent amount of Lucifer yellow did not exhibit displaying these physiological characteristics consistently la- significant levels of fluorescence when examined with the rho- beled large (30 pm diameter or greater) multipolar neurons lo- damine filter, and (2) the regions of double-labeled neurons that 3466 Grace and Onn l identification and Characterization of DA Neurons in vitro

Figure 1. Basedon their electrophysiologicalcharacteristics, at least two neuronal types located in the DA-containing regionsof the rat midbrain slice could be distinguished.One neuron type could be identified by its short duration spikesand its firing pattern. At resting membranepotentials, this cell type exhibited repetitive firing of action potentials in responseto a depolarizing pulse (A). On the other hand, hyperpolarization of the membraneof this neuron causedthe depolarizing pulse to trigger a low threshold depolarization and a rebound burst of action potentials (B). Neuronswith theseelectrophysiological characteristics did not show the presenceof tyrosine hydroxylase-like immunoreactivity in double-labeling studiesand hencewere not representativeof the DA-containing neuronsin this region. C, The membranepotential changesunderlying spontaneous action potential generation in the other classof midbrain neurons recordedin vitro were much like those observedfor identified DA neurons in vivo.Spike activity wasdriven by a slow depolarization,depolarizing the membranefrom its restingstate (dashed line 2) to that ofits characteristically high spikethreshold (dashed line l), in this case,approximately -33 mV. The action potentials werefollowed by a prominent afterhyperpolarization (dashedline 3), which hyperpolarizedthe neuron below resting potential and decayedinto the onsetof the next slow depolarization. Neuronswith this electrophysiologicalpattern were consistently labeled for tyrosine hydroxylase-like immunoreactivity (Fig. 2).

showed the highest level of Lucifer yellow fluorescence usually ventral tegmental area, which had 3-5 major dendrites pro- were distinct from those parts of the neuron that showed the jecting radially from the soma. The thin (0.5 pm diameter) axon most intense levels of rhodamine fluorescence (Fig. 2). If the typically arose from a major dendrite or somatic appendage fluorescence observed under the rhodamine filter was due to (Fig. 4A) and, for substantia nigra neurons, traveled in a medio- bleedthrough from the Lucifer yellow stain, then the brightest ventral direction within the zona compacta and along the border areas of the cell under each filter setting would be identical. between the zona compacta and zone reticulata before turning in an anterior direction at the midline. Dendrites typically ex- DA neuron morphology hibited spine-like protrusions that occurred at irregular inter- The morphological characteristics of DA neurons were studied vals, ranged in length from 5 to 50 pm, and were 0.5-2 pm in by injecting the highly fluorescent dye Lucifer yellow into neu- diameter (Fig. 4B). These spinules were particularly prominent rons displaying the characteristic electrophysiological properties on the dendrites of ventral tegmental DA neurons. In each case outlined for DA neuron identification. Morphological analyses where the dendrites of DA neurons could be followed to their were carried out on 2 1 DA neurons injected with Lucifer yellow. termination, they were observed to end in a tuft of 2-4 fine The labeled neurons showed a distinct morphology, consisting (0.2-0.8 pm diameter) recursive processes that were seen to of medium-sized somata that gave rise to between 2 and 6 thick encompass a circumscribed region of tissue (Fig. 4C). In general, major dendrites. These sparsely branching dendrites typically this morphology is consistent with morphological studies of exhibited their first bifurcation at distances of 50 pm or more identified dopaminergic neurons in these brain regions (Rinvik from the soma and extended for distances of up to 1200 Wm. and Grofova, 1970; Gulley and Wood, 197 1; Schwyn and Fox, From this sample of injected cells, 3 morphological types of DA 1974; Lindvall and Bjorklund, 1974, 1978; Hijkfelt et al., 1976; neurons could be distinguished (Fig. 3): (1) multipolar neurons Juraska et al., 1977; Groves and Linder, 1983). (20-35 pm diameter) in the ventral zona compacta, which had 2-4 major dendrites extending in medial and lateral directions, DA neuron spike generation in addition to l-3 dendrites extending ventrally into the zona Most DA neurons recorded intracellularly in vitro were spon- reticulata, (2) fusiform neurons (15-25 pm diameter) in the taneously active, with firing rates ranging between 1 and 7 Hz dorsal portion of the zona compacta, which had 2-5 major (average = 3 f 2.5 Hz; n = 52). Spontaneous action potential dendrites emanating from the soma and extending in medial discharge was preceded and apparently driven by a slow de- and lateral directions but remaining within the confines of the polarization (amplitude = 22 + 3 mV, duration = 295 + 84 zona compacta, and (3) fusiform and multipolar neurons in the msec, depending on firing frequency; n = 35). The slow depo- The Journal of Neuroscience, October 1989, 9(10) 3467

Figure 2. One class of midbrain neurons could be distinguished electrophysiologically based on the presence of a slow depolarization and a pacemaker-like firing pattern. These cells consistently demonstrated the presence of tyrosine hydroxylase-like immunoreactivity, as shown by labeling cells with these unique physiological characteristics by intracellular injection of the dye Lucifer yellow, followed by immunocytochemical localization of tyrosine hydroxylase-like immunoreactivity using antibodies tagged with a rhodamine (red fluorescent) label. A, A multipolar neuron in the substantia nigra injected with Lucifer yellow (arrow, A,) also demonstrated the presence of tyrosine hydroxylase-like immunoreactivity (arrow, A,) B, A bipolar neuron in the ventral tegmental region that exhibited electrophysiological characteristics similar to those described above was also stained with Lucifer yellow (arrow, B,) and demonstrated tyrosine hydroxylase-like immunoreactivity (arrow, B,). Scale bar, 50 pm.

larization may play a significant role in maintaining spontaneous because of the absence of a stable membrane potential in these action potential generation in DA neurons, since it depolarizes pacemaker-like neurons but was estimated as the inflection in the membrane potential from its resting levels (-57 f 4 mV; the membrane potential occurring between the offset of the af- n = 35; range: -52 to -67 mV) to the depolarized spike thresh- terhyperpolarization and the initiation of the next slow depo- old of DA neurons (Fig. lc). The resting potential of sponta- larization. Spontaneously occurring action potentials were trig- neously firing DA neurons was difficult to establish precisely gered at membrane potentials averaging -36 f 4 mV as ,

’ _ _.. \,,* , .- -.. ,; , i The Journal of Neuroscience, October 1999, 9(10) 3469 measured at the soma (n = 74), which is similar to that of DA neurons recorded in vivo (Grace and Bunney, 1983a). Action Firing pattern potentials were characteristically long in duration, with an av- A second characteristic difference in the physiology of DA neu- erage spike duration of 2.7 f 0.5 msec (range: 1.8-3.5 msec; n rons recorded in vitro versus those described for DA cells in = 8 1). The amplitudes of the DA neuron action potentials varied vivo pertains to their fning pattern. DA neurons recorded in vivo with membrane potential, with an average spike height of 77 fired in either an irregular single spiking mode or, on depolar- k 4 mV (range = 65-92 mV, II = 70) measured from the spike ization, in a burst firing pattern (Bunney et al., 1973; Grace and peak to the beginning of the afterhyperpolarization. This type Bunney, 1984a, b). In contrast, DA cells recorded in vitro fired of measurement was necessary to circumvent the problems of exclusively in a highly regular pacemaker-like pattern. Depo- measuring spike height using the pre-spike membrane potential larization of DA neurons in vitro resulted in an increased rate- as baseline because of the presence of the slow depolarization of-rise of the slow depolarization and, hence, an increased firing preceding the spike. DA neuron action potentials were followed rate but did not alter the very regular pacemaker firing pattern by a prominent afterhyperpolarization, which hyperpolarized (Fig. 5@. The state of the electrode balance was monitored the cell before the onset of the next slow depolarization. during polarization of the membrane to ensure that the mem- Although many of the characteristic electrophysiological brane potential observed on the oscilloscope accurately reflected properties of DA neurons recorded in vitro were similar to those the membrane potential of the DA cell soma. Burst firing was observed during in vivo recordings from identified DA neurons, not observed in any of the identified DA neurons recorded in 2 significant differences in their physiology were noted: (1) DA vitro. Furthermore, a firing pattern resembling the burst firing neurons recorded in vitro had much higher input resistances, pattern described for DA neurons recorded in vivo (Grace and and (2) unlike the irregular single spiking or burst firing patterns Bunney, 1984b) or for putative DA neurons recorded in vitro reported for spontaneously firing DA neurons recorded in vivo (Nedergaard et al., 1988a, b) could not be induced in identified (Grace and Bunney, 1984a, b), DA cells recorded in vitro only DA neurons in this preparation by any of the current injection fired spikes in a very regular, pacemaker-like firing pattern. and membrane polarization parameters examined in this study. Input resistance TTX-sensitive membranepotentials The input resistance of the DA cells recorded was determined Slow depolarization. In spontaneously firing DA neurons, the by plotting the amplitudes of the hyperpolarizing current pulses slow depolarization is observed as a slowly developing depo- injected into the cell against the resultant membrane voltage larization of the membrane from resting potentials. The slow deflections produced. The input resistance measured in DA neu- depolarization averaged 22 mV f 3 mV (n = 35) which was rons was found to be nonlinear over the range of current injec- sufficient to depolarize the DA neuron membrane from resting tion amplitudes tested. The peak input resistance measured in potentials to the comparatively high spike threshold found for the linear region of the resistance plot (i.e., near the resting these neurons in vitro (i.e., -36 mV) as well as in vivo (-41 potential) was 168 + 61 MQ (Fig. 5A: range: 80-320 MO, n = mV; Grace and Bunney, 1984a). The slow depolarization can 38), which was about 5 times higher than that reported for DA be triggered alone by small depolarizations of the DA neuron. neurons recorded in vivo (i.e., 3 1 f 7.4 MQ, range = 18-45 MQ) Larger depolarizations caused the slow depolarization to further (Grace and Bunney, 1980, 1983a). During intracellular injection depolarize the cell sufficiently to trigger an action potential (Fig. of long (e.g., > 200 msec) hyperpolarizing current pulses, a slow- 6A, B). The slow depolarization can also be triggered as a com- ly developing sag in the membrane voltage was observed. This ponent of the rebound depolarization occurring at the termi- is illustrated as a decrease in the slope of the current/voltage nation of brief membrane hyperpolarizations. Thus, injection plot at long time points (i.e., a decrease in the steady-state input of small hyperpolarizing current pulses into quiescent DA neu- resistance) below that corresponding to the peak input resistance rons will cause a rebound depolarization, which can then trigger measured. Plotting both the peak and steady-state input resis- spike discharge (Fig. 60,). The slow depolarization also can be tances showed that these measures deviated from linearity as inactivated by briefly hyperpolarizing the membrane. Thus, in- well as from each other (Fig. 5A). This apparent decrease in jection of short-duration hyperpolarizing current pulses during input resistance during hyperpolarization ofthe membrane shows a spontaneous slow depolarization resets the membrane poten- that this event is similar to the anomalous rectifier described tial to baseline levels (Grace and Bunney, 1984a), as will the previously (Katz, 1949; Adrian, 1969). AHP following an action potential (Figs. lC, 6B). The duration

t

Figure 3. The morphology of DA neurons was analyzedby injecting physiologically identified midbrain DA neurons with the highly fluorescent dye Lucifer yellow. The labeled cells had medium-sized somatathat gave rise to 2-6 sparselybranching dendrites. These cells could be divided into 3 subclassesbased on their location and morphology.A, DA neuronsin the ventral regionsof the substantianigra zona compactahad multipolar somata.This cell had 4 major dendrites extending laterally from the soma, which remained within the zona compacta.In addition, 1 dendrite extendedventrally into regions deep within the zona reticulata. B, DA neurons in the more dorsal regions of the substantia nigra had fusiform somata.This neuron had 5 major dendrites extending mediolaterally from the soma,all of which remainedentirely within the confinesof the zona compacta.C, The cell body of this ventral tegmentalDA neuron was multipolar in shape,with 5 major dendrites extending from the soma.The dendrites typically projectedin a radial array from the soma in this region, apparently dependingon the orientation of the DA cell body within the tegmentum.Midline is to the right in (A) and (B) and to the left in (C’).Scale bars, 75 pm in A and C, and 50 pm in B. 3470 Grace and Onn * Identification and Characterization of DA Neurons in vitro

f

Figure 4. DA neurons labeled by intracellular injection of Lucifer yellow exhibited several unique morphological characteristics. A, In each case where this structure could be identified, the thin, 0.5 Frn diameter axon (arrows) was observed to arise from a major dendrite or a somatic appendage rather than directly from the soma. B, DA neuron dendrites exhibited infrequent spine-like protrusions that extended 3-l 5 Frn from the dendritic shaft (arrows). These protrusions were particularly prominent in labeled DA neurons in the ventral tegmental area. C, In each case where the dendrite could be followed to its termination site within the plane of the slice, the dendrite ended in a fork-like tuft. This consisted of recurrent, thin processes that appeared to surround a section of the underlying midbrain, possibly forming a basket around non-DA cells in this region. Many of the fine recurrent processes in this figure are not within the plane of focus of the primary dendrite. Scale bar, 20 pm. The Journal of Neuroscience, October 1989, 9(10) 3471 of the slow depolarization often will outlast the depolarizing A stimulus that triggered it. Thus, short-duration depolarizing nA I” pulses will still elicit a slow depolarization and an action po- -0.30 -0.25 -0.20 -0.15 -0.10 -0.05 tential in DA neurons even if the pulse is terminated before the I spike threshold is attained (Fig. 6c). These findings provide additional evidence that this pacemaker-like depolarization ex- hibits voltage-dependent properties (Grace and Bunney, 1984a). The slow depolarization is strongly influenced by the temper- ature of the slice preparation, with temperatures below 37°C decreasing the rate of activation of the slow depolarization. Indeed, very few spontaneously firing DA neurons could be recorded when the temperature of the preparation fell below 35°C. 0 R-200MCl / The slow depolarization component of the rebound depolar- 1’ -40 ization triggered in DA cells cannot be attenuated by admin- . /’ istering calcium blockers. Thus, the addition of the calcium / / blocker cobalt (2 mM) to the Ringer’s (containing 1.0mM calcium / -50 / in Tris buffer in place of the bicarbonate buffer) blocked acti- / vation of the rebound spike following a hyperpolarizing current pulse but failed to attenuate the entire rebound response (Fig. 60,,,). After cobalt administration, the amplitude of the control rebound depolarization could not be restored regardless of the amount of conditioning hyperpolarization used. Subsequent administration of the sodium blocker TTX (1 x 10e6M final concentration) to the Ringer’s blocked activation of the re- maining cobalt-insensitive rebound depolarization (Fig. 60,). Thus, blocking calcium channels with cobalt revealed an iso- lated rebound slow depolarization that could be blocked by TTX. Furthermore, blockade of this depolarization with TTX also eliminated the spontaneous or depolarization-induced membrane potential oscillations underlying the pacemaker fir- ing pattern. 1 SW High threshold sodium spikes. Blockade of calcium channels Figure 5. A, Input resistanceof DA neurons recordedin vitro. The by the admistration of l-2 mM cobalt chloride into the Ringer’s input resistanceof this nonfiring DA neuron was calculatedby plotting caused a gradual decrease in the amplitude of the action poten- the peak membranevoltage deflections (solid circles) and the steady- tial and afterhyperpolarization, which culminated in the block- state voltage deflections from resting potential (open circles) obtained ade of spontaneous and depolarization-induced action potential during the injection of constantcurrent hyperpolarizingpulses (see inset). firing (Fig. 7, A, B). Thus, after cobalt treatment, depolarizations DA neurons had significantly higher input resistancesin the in vitro preparation than observedin vivo. In this case,the input resistancewas similar to those eliciting action potentials in untreated neurons 200 Mo. Both the instantaneousand time-dependentcomponents of were ineffective in eliciting active spiking. However, much larger the anomalous rectification were observed,as shown by the deviation depolarizations resulted in the triggering of large amplitude (42 of the peak input resistanceregression line from linearity and the de- ? 15 mV; n = 19; range: 20-60 mV) cobalt-insensitive spikes flection of the steadystate from the peakinput resistances,respectively. B, Firing pattern of DA neurons recordedin vitro. Unlike DA neurons (Fig. 7C). These spikes were characterized by their fast time recorded in vivo, identified DA cells in the in vitro preparation fired course (0.9 -t 0.2 msec duration; n = 19; range: 0.5-1.3 msec) exclusively in a highly regular, pacemakerpattern, and demonstrated and their high threshold (-33 & 15 mV; n = 23; range: - 15 little variation in interspike interval over time. Furthermore, depolar- to -60 mV). Both the threshold and amplitude of these fast ization (top 2 traces,0.05 nA and 0.1 nA) or hyperpolarization (bottom) spikes were highly dependent on the membrane potential, as trace, -0.07 nA) of the DA neurons from baseline (third trace from top) only changedthe frequencyof spike dischargewithout altering the reflected in the comparatively large SDS reported for measures firing pattern of the DA neuron. of mean spike amplitude and threshold. Furthermore, although DA neuron action potentials generally exhibit a minimum in- terspike interval of >40 msec, larger depolarizations of cobalt- DA neuron action potential, the fast spikes were not followed treated DA neurons elicited these fast spikes at comparatively by large amplitude, long-duration afterhyperpolarizations. In- high frequencies (e.g., 10-l 5 msec interspike intervals, Fig. 70. deed, these spikes presented little evidence of a calcium-acti- Without cobalt administration, depolarizing DA neurons from vated afterhyperpolarization. A similar low-amplitude, fast spike resting potentials could not trigger isolated cobalt-resistant that fired without triggering a prominent afterhyperpolarization spikes-instead, only full amplitude (i.e., >70 mV) action po- was observed in vivo during antidromic activation of DA neu- tentials were elicited. However, similar fast spikes could be rons (Grace and Bunney, 1983b). This spike was shown to be triggered in DA neurons in normal Ringer’s following large hy- the initial segment (IS; Coombs et al., 1957) spike component perpolarizations. In these cases, the fast spikes were triggered of the DA cell action potential (Grace and Bunney, 1983b). The at much more hyperpolarized spike thresholds (Fig. 7E). Both fast spike component observed here is similar in time course the slow depolarization and the fast spikes were blocked by and amplitude to this previously described IS spike. In addition, administering TTX (Fig. 7P). In contrast to the full amplitude the fast spike triggered during rebound depolarizations had ap- 3472 Grace and Onn - Identification and Characterization of DA Neurons in vitro

Dt Control

mV

DO Cobalt -201

D3 Cobalt + TTX

I -60 o nA -,

Figure 6. Evidencesuggests that the slow depolarization may exhibit voltage-dependentproperties. In a hyperpolarizedDA neuron, the injection of depolarizing current activates the slow depolarization (A), whereasslightly larger depolarizationstrigger an action potential (B; dashedlines = depolarization plateau before the slow depolarization). The slow depolarization, once activated, often outlasts the duration of the depolarizing event, being sufficiently regenerativein nature to trigger a spike after the termination of the depolarizing current pulse (C). The slow depolarization can be distinguishedfrom the calcium-dependentrebound depolarization by its sensitivity to blockers.D,, Responseof a hyperpolarizedDA neuron to intracellular injection of a hyperpolarizing current pulse. At the offset of current injection, at least two membraneevents occur: (1) a delay in the return of the membraneto its baselinemembrane potential (dashedline) and (2) a depolarization of the membranebeyond its resting potential, triggeringan action potential. D,, Administration of the calcium blocker cobalt (2 mM) to the Tris-buffered Ringer’ssolution attenuatesthe rebound depolarization and prevents activation of spike discharge.D,, Administration of the sodium channel blocker TTX (1 PM) blocks this remaining rebound depolarization (i.e., the TTX-sensitive slow depolarization). After a transient delay, the DA cell membranerepolarizes to its baseline membranepotential without a concomitant rebound depolarization.

proximately the same amplitude and time course as the notch TTX-sensitive depolarization. After selective blockade of the in the rising phase of the DA neuron action potential. slow depolarization by TTX, the remaining TTX-insensitive depolarizing the DA cell membrane. Increasing the membrane Cobalt-sensitivemembrane potentials blocker cobalt (2 mM) into the Ringer’s (Fig. 8A). Low threshold depolarization. A rebound response that tran- High threshold calcium spikes. Following TTX administra- siently depolarizes the DA neuron above its steady-state, base- tion, only the low threshold depolarization can be activated by line membrane potential can be observed following a brief hy- depolarizing the DA cell membrane. Increasing the membrane perpolarization of the DA neuron membrane or by injecting depolarization also failed to trigger other depolarizing mem- subthreshold levels of depolarizing current into hyperpolarized brane processes in TTX-treated DA neurons, even if compar- DA neurons. This rebound depolarization appears to be com- atively large depolarizations were used (Fig. 8B). In contrast, posed of more than a single component. The slow depolarization after the application of the potassium channel blocker TEA (2- was shown to be one element of this response, since a portion 5 mM) to TTX-poisoned slices, one or more large amplitude of the resultant depolarization can be attenuated by adminis- (75 + 11 mV, n = 33; range: 61-85 mV), long duration (12.2 tering TTX to the slice. However, blockade of the sodium- f 3.4 msec; range: 7-20 msec; IZ = 36) TTX-insensitive spikes mediated slow depolarization and spike generation by the ap- could be elicited with only moderate depolarizations (Fig. 8B,). plication of TTX also reveals the presence of a TTX-insensitive These spikes were most likely calcium-mediated, since they had low threshold depolarization. As shown above for the TTX- a comparatively slow time course with respect to that of the sensitive component of the rebound response, this TTX-insen- sodium spikes and could be blocked by administering cobalt to sitive depolarization can be triggered either by terminating the the Ringer’s (1 mM cobalt: 1 mM calcium in Tris-buffered Ring- hyperpolarization of the DA cell membrane or by directly de- er’s, Fig. 8B,). Unlike the low threshold depolarization, depo- polarizing a hyperpolarized DA cell. However, depolarization larization of the DA neuron facilitated current-induced trigger- of a DA cell already in a depolarized state will not activate this ing ofthis spike. Thus, in terms of threshold, duration, amplitude, The Journal of Neuroscience, October 1989, 9(10) 3473

A Control c Cobalt 20 20

0 0 mv -20 -40 / Figure 7. Effectsof calcium blockade -60 0.5 mv -20 L- I 1 on action potential generation in DA o nA neurons.A, Depolarization of a slowly L I-0.5 D Cobalt + TTX firing DA neuron triggersthe discharge B Cobalt of a seriesof action potentials.B, After cobalt (2 mM) is addedto the Tris-buff- 10 eredsuperfusion medium, the depolar- ization is no longer capableof causing action potential discharge.C, Spikefir- ing can be achieved with larger depo- larizations. The spikesproduced exhib- it high thresholds,are short in duration (l-l.5 msec),and are not followed by 0 0.5 t.0 15 2.0 afterhyperpolarizations of significant 0 0.5 I.0 15 seconds amplitude or duration. The absenceof seconds large afterhyperpolarizationsfollowing E Control fast spikesallows large membranede- polarizations to trigger a seriesof these -30 fast spikes in cobalt-treatedDA neu- -50 rons. The spikes are elicited at inter- mV-70 l--l spike intervals that are much shorter than thoseobserved for action potential firing in DA neurons in normal Ring- l- er’s. D, Although this fast spike is not blocked by administering calcium F TTX channel blockers,administration of the sodium blocker TTX (24 PM) sup- -2oj pressesthe dischargeof this fast spike, supportingits tentative identification as a sodium-dependentIS spike.E, Large membranehyperpolarizations can elic- it a rebound spike showing a similar -12OJ V amplitude and time coursewithout co- -/m-m-... P-Y 1’0 nA balt pretreatment.F, The fast rebound -I spike can be blocked by administering 0 200 400 600 800 100.3 12oa msec TTX (2-4 PM) into the Ringer’s. and ionic dependency, these TTX-insensitive spikes were sim- of the anomalous rectifier and (2) a delay in repolarization fol- ilar to the high threshold spikes (HTS) described in other prep- lowing a hyperpolarizing current pulse (henceforth referred to arations. as the delayed repolarization). The anomalous rectifier was in- As shown above, HTSs could be triggered after the admin- troduced previously in the section on input resistanceto account istration of TEA by depolarizations that previously were incapa- for the sag. The delayed repolarization occurring during mem- ble of triggering large amplitude, TTX-resistant spikes. TEA brane repolarization apparently is inactive at resting potentials administration, therefore, lowered the apparent threshold of the but is activated during the transition of the membrane potential HTS, at least as measured by electrodes located in the soma. from a more hyperpolarized state to a depolarized level, i.e., at Thus, after TEA administration, the HTSs could be triggered the offset of membrane hyperpolarization. This delayed repo- by small membrane depolarizations (Fig. 94, as well as by the larization was observed to slow the repolarization of the mem- rebound depolarization that occurs following small hyperpo- brane potential to baseline levels (Fig. 10B). Although the cur- larizations (Fig. 9B). In hyperpolarized DA neurons, membrane rent underlying this event could not be identified directly without depolarizations that were subthreshold for activating the HTS voltage clamping, this event has several properties in common were still capable oftriggering the low threshold, cobalt-sensitive with the A current (IJ described in other preparations, such as depolarization. In contrast to the cobalt-insensitive IS spike activation during membrane repolarization following a hyper- reviewed above, the HTSs were followed by large amplitude, polarizing pulse, failure to activate at membrane potentials be- long duration afterhyperpolarizations, with both the amplitude low - 60 mV (Fig. 1OA), and an insensitivity to TEA (Fig. 1OB), and duration of the afterhyperpolarization dependent on the as described by Connor and Stevens (197 1). However, in this number of HTSs elicited (Fig. SC’). preparation, administration of the selective potassium blocker 4-AP did not block this putative I, (Fig. lOC), in contrast to Nonregenerativemembrane potentials the pharmacological specificity of 4-AP in blocking analogous Long duration (i.e., >300 msec) hyperpolarizations of DA neu- currents in other preparations (Connor and Stevens, 197 1; Gus- rons revealed the presence of two voltage-dependent conduc- tafsson et al., 1982). Cobalt also was ineffective in measurably tances: (1) a voltage-dependent sag that occurred during mem- attenuating this delayed repolarization. brane hyperpolarization and exhibited properties similar to those A voltage-dependent delay in membrane repolarization could 3474 Grace and Onn l Identification and Characterization of DA Neurons in vifro

Figure 8. A, The low threshold de- polarization demonstrates a markedly slower activation and a different sen- sitivity to ion channel blockers than found for the slow depolarization. A,, Control B, TTX Continuous hyperpolarization of the DA AI neuron membrane prevents sponta- 10 j neous spike discharge in this cell. In- O- I jection of a pulse of depolarizing cur- mV u rent leads to membrane depolarization -50.- and spike firing. A,, Administration of mV the sodium channel blocker TTX (1 PM) to the superfusion medium blocks spike firing and reveals a TTX-insensitive component of the membrane depolar- ization. - A,, Subsequent administra- tion ofcobalt (2 mM) to the T&-buffered B, TTX + TEA superfusion medium blocked this low A2 TTX threshold depolarization. Current in- jection parameters are identical in each -30 -40 case. B, In addition to the slower, low 1 threshold cobalt-sensitive depolariza- tions observed in DA neurons after TTX “:::P mV administration, a type of high thresh- -70' old, fast calcium spike can also be elic- ited. B,, Increasing levels of depolariza- A, TTX + Cobalt tion fails to elicit spiking after TTX administration. B,, As TEA (5 mM) is -301 - I administered to the slice, it initially mV causes an enhancement of the low threshold depolarization occurring in response to similar levels of membrane 1 83 TTX + TEA + Cobalt depolarization. Over time, this low o nA threshold depolarization increases in -, amplitude until it triggers an all-or- nothing, large amplitude spike. The low 0 200 400 600 msec threshold depolarization was not re- placed by the spike, since the low threshold depolarization can still be triggered alone with smaller depolar- 1 0 nA izing stimuli. B,, Subsequent applica- _I t tion of cobalt to the Tris-buffered Ring- -1 er’s (2 mM) blocks activation ofthe high 0 200 400 600 800 lcco threshold, TTX-insensitive spike. msec

act to modulate other rebound events, such as the rebound depolarizations. In this preparation, hyperpolarizations suffi- Neuron identification cient to activate this delayed repolarization cause an attenuation Based on electrophysiological criteria, 2 general classes of neu- of the rebound depolarization (Fig. 11A) described earlier in rons could be distinguished in the zona compacta region of the this paper. The degree of attenuation produced appears to de- substantia nigra and in the ventral tegmental area. The principal pend on the amount of membrane hyperpolarization induced. neuron encountered was typically spontaneously active and fired Thus, although a small hyperpolarization of the DA neuron will in a pacemaker pattern. The long duration, high threshold action trigger a rebound membrane depolarization and spike firing, potentials were preceded by a slow depolarization and followed increasing levels of hyperpolarization will actually prevent this by a prominent afterhyperpolarization. Labeled neurons of this rebound excitation at the same time that it activates the delayed type had medium-sized fusiform or multipolar somata that gave repolarization (Fig. 1 le). rise to a small number of sparsely branching dendrites. These neurons consistently demonstrated double labeling for tyrosine Discussion hydroxylase-like immunoreactivity, indicating that they were Identifying the factors that regulate DA neuron activity and how of the dopaminergic subtype. A second neuron type rarely ex- this regulation is affected by drug administration may yield hibited spontaneous activity and fired short duration (< 1.5 msec) information of relevance to the etiology and treatment of some action potentials in a repetitive manner when depolarized from neurological and psychiatric disorders. However, with respect resting potentials but would fire high frequency (3-5 msec in- to applying the data gathered from recordings of neuronal ac- terspike interval) bursts of spikes when depolarized from hy- tivity to the analysis of drug action, it is first necessary to es- perpolarized membrane potentials. Labeled neurons exhibiting tablish that the neurons recorded are indeed of the neurochem- these electrophysiological properties had large somata (> 30 pm ical subtype one wishes to investigate. diameter), with numerous highly branched dendritic processes The Journal of Neuroscience, October 1989, 9(10) 3475

A 251

0

m” -25

-60 -50

i( -751 1 1 I 0 nA t -I mV B Figure 9. The TTX-insensitive spike 25. cannot be triggered in TTX-treated DA -60 neurons without administering TEA. O- However, after treatment with TEA, this spike can be triggered easily by small t-n" -25. 0 membrane potential changes. A, Fol- lowing treatment with TTX and TEA, -50-A -20 depolarization of a depolarized DA v- mV neuron triggers a calcium spike. B, This -40 spike also can be triggered by the re- -75 1 1 bound depolarization occurring at the 0 nA -60 offset of a brief hyperpolarization of the -1 membrane of depolarized DA neurons. 0 100 200 300 400 500 600 C, The TTX-insensitive spikes are fol- msec lowed by prominent afterhyperpolar- izations. In this case, equal amplitude depolarizations of a DA neuron treated with TTX and TEA triggers 1,2, or 3 spikes (C,-C,). Overlaying these 3 traces shows that the amplitude and duration of the afterhyperpolarizations following 0.5 the current pulses are dependent on the ------I 0 nA number of spikes elicited, with more -0.5 spikes associated with more prominent 0 0.5 1 .o 1.5 afterhyperpolarizations. (C,, overlay of seconds traces C,-C,.)

B Control

-60 mV mvIti-;-

-00

c 4-AP

mv -60

0 0.5 1.0 1.5 seconds r 1 7 I t 0 nA -1

0 0.5 1.0 1.5 seconds Figure IO. A, Both the anomalous rectifier and the delayed repolarization exhibit voltage-dependent properties. In a DA neuron showing low levels of spontaneous activity, depolarization decreased and hyperpolarization increased the time-dependent “sag” in the membrane potential, which is reported to reflect the magnitude of the anomalous rectification. Hyperpolarization of the membrane below about - 60 mV also prevented activation of the delayed repolarization. B, C, Although the delayed repolarization induced in DA neurons during repolarization from hyperpolarized potentials shows a time course and voltage dependence similar to that described for I,-type currents, it is not blocked to a significant extent by 4-AP. Thus, application of 4-AP to a TEA- and TTX-treated DA neuron is capable of blocking only a portion of this delayed repolarization. B, Control response, TTX = 1 PM; TEA = 4 mM, C, Overlay of control response (I) with response obtained after addition of 4-AP. Concentrations: 5 mM (2) and 20 mM (3). 3476 Grace and Onn - Identification and Characterization of DA Neurons in vitro

DA neuron morphology Consistent with the morphology reported for physiologically identified DA neurons labeled during in vivo intracellular re- cordings (Grace and Bunney, 1983b, Tepper et al., 1987) and with Golgi (Schwyn and Fox, 1974; Juraska et al., 1977) fluo- rescence histochemical (Lindvall and Bjiirklund, 1974, 1978), t 0 nA and immunocytochemical (Hokfelt et al., 1976) studies of mid- -0.5 brain DA neurons, the identified DA neurons labeled here could 0 200 400 600 800 be divided into 3 morphological classes: (1) multipolar sub- msec stantia nigra cells, (2) bipolar substantia nigra cells, and (3) ventral tegmental area neurons that ranged from multipolar to B bipolar in shape. The DA neurons exhibited 3-6 thick, sparsely branching major dendrites emanating from the soma, with the dendrites typically oriented in a mediolateral direction and re- maining within the the dorsal-ventral confines of the substantia nigra zona compacta or extending ventrally into the zona retic- -20 ulata (Juraska et al., 1977; Preston et al., 198 1; Grace and Bun- mV I ney, 1983b, Tepper et al., 1987; Grace, 1987, 1988). The den- drites of the bipolar substantia nigra neurons typically remained within the confines of the zona compacta, which is consistent with the morphology described for the dorsal tier DA neurons (Fallon et al., 1978; Gerfen et al., 1987a, b) and the acetylcho- linesterase-rich DA neurons identified by Jiminez-Castellanos I 0 nA and Graybiel in primates (1987). The more ventral multipolar t -2.5 DA neurons of the substantia nigra also had dendrites confined 0 100 200 300 400 500 to the zona compacta but, in addition, had l-3 dendrites that msec descended into the zona reticulata. This morphology is similar to that ofthe ventral tier DA neurons described by others (Bjork- Figure II. A, The amplitude of the rebound depolarization in DA lund and Lindvall, 1975; Gerfen et al., 1987a, b). Based on this neurons is strongly influenced by the baseline membrane potential. In- dendritic branching pattern, the more dorsal fusiform DA neu- tracellular injection of a hyperpolarizing current pulse into a slightly depolarized but nonfiring DA neuron (upper trace) triggers a rebound rons would be restricted to sampling synaptic input confined to depolarization at the offset of the pulse. However, injection of a pulse the zona compacta region, whereas the more ventral multipolar of current of the same amplitude into a hyperpolarized DA neuron neurons also have dendrites that could receive afferent inputs (lower trace) does not trigger a rebound depolarizing response. B, In terminating in the zona reticulata as well. DA neuron dendrites DA neurons, larger membrane hyperpolarizations produce compara- were found to exhibit spinule-like protrusions at irregular in- tively smaller rebound depolarizations. Thus, hyperpolarization often triggers rebound spiking in nonfiring DA neurons (upper trace). In con- tervals (Juraska et al., 1977; Preston et al., 198 1; Tepper et al., trast, larger hyperpolarizations prevent rebound activation of the low 1987; Grace, 1987, 1988), with the dendrites ending in forks or threshold depolarization, presumably because of the concurrent acti- tufts (Preston et al., 1981; Grace and Bunney, 1983b). These vation of the delayed repolarization (lower trace). (Dashed lines = base- terminal dendritic forks consisted of 2-3 fine processes that line membrane potential.) appeared to “wrap around” an element in the zona reticulata or in the zona compacta. Indeed, this morphological special- emanating from the soma, some of which were observed to ization is not dissimilar from that described for the basket-like extend dorsally into the reticular formation. Neurons of this terminations of axons around the base of pyramidal neurons, type did not demonstrate tyrosine hydroxylase immunoreactiv- as described in the cerebral cortex (Jones, 1975). One speculative ity. hypothesis to account for this type of morphological arrange- Drawing from the range of criteria used to identify DA neu- ment would be that the terminal tufts are dendritic release sites rons in vivo, this second neuronal type had sufficiently different for DA acting on non-dopaminergic nigral neurons, such as the morphological and physiological characteristics to allow it to be DA-sensitive nigrothalamic neurons reported by others (Ruf- readily distinguished from DA neurons. Thus, in addition to fieux and Schultz, 1980; Waszczak and Walters, 1983). the inability to label this neuron immunocytochemically, this latter neuronal type resembled the morphological characteristics Physiological properties of DA neurons of neurons that had been classified as non-dopaminergic based Despite the presence of 3 morphological subtypes of DA neu- on in vivo (Tepper et al., 1987: non-DA neurons) and in vitro rons, all identified DA neurons exhibited similar electrophys- (Nakanishi et al., 1987: zona reticulata neurons) recordings and iological characteristics. Furthermore, DA neurons recorded in staining of neurons in this brain region. Furthermore, the firing vitro displayed many of the same physiological characteristics pattern observed for this class of neurons is consistent with the reported for in vivo intracellular recordings from this class of electrophysiological characteristics reported by others for neu- cells, such as (1) a long duration action potential, (2) a slow, rons in this brain region (Llinbs et al., 1984; Nedergaard et al., depolarizing pacemaker potential, (3) a high spike threshold, 1988a, b; Greenfield et al., 1988), which some have identified and (4) a prominent afterhyperpolarization (cf. Grace, 1987). as non-dopaminergic zona reticulata neurons (Nakanishi et al., However, 2 striking differences were found in the membrane 1987). properties of DA neurons recorded in the in vitro preparation The Journal of Neuroscience, October 1989, 9(10) 3477 when compared to those reported in vivo. These differences though intracellular injection ofthe calcium chelator EGTA into relate to DA cell input resistance and firing pattern. DA cells DA neurons in vivo will cause them to fire in a very similar, recorded in vivo typically exhibit input resistances averaging highly regular pacemaker pattern (Grace and Bunney, 1984a) about 30 MQ (Grace and Bunney, 1983a). In contrast, DA cells as well as prevent membrane depolarization from eliciting burst recorded in vitro have input resistances 3-6 times this value; firing (Grace and Bunney, 1984b). The relevance of these ob- i.e., averaging about 170 MB, which is consistent with that servations to the data discussed here, however, is not clear at reported by others using in vitro recordings from midbrain slices present. (Pinnock, 1984; Kita et al., 1986; Lacey et al., 1987; Silva and Bunney, 1988) and dissociated cell cultures (Chiodo and Ka- Anomalous rectification patos, 1987; Chiodo, 1988; Ort et al., 1988). Although it is Intracellular injection of hyperpolarizing current pulses elicited possible that this may represent a difference in the stability of 2 events that may play a role in regulating DA cell activity. the electrode penetration between these two preparations, this Hyperpolarization of DA neurons to -75 mV or more was explanation appears to be inadequate to account completely for accompanied by the development of a prominent sag in the this disparity, since (1) DA neurons recorded intracellularly in membrane potential (Grace and Bunney, 1983a; Pinnock, 1985; vivo exhibited action potential durations, firing rates, and firing Kita et al., 1986), thereby decreasing the amount of hyperpo- patterns similar to those observed extracellularly in vivo where larization produced. The change in conductance underlying this penetration injury should not be a factor, and (2) penetrations sag has properties similar to those described for the anomalous of spontaneously firing DA neurons in vivo could be maintained rectifier in cortical neurons (Schwindt et al., 1988) and appears for periods of up to 2.5 hr without significant increases in input to be larger in amplitude than that observed for DA neurons resistance (Grace, 1987). The observation that DA neurons re- recorded in vivo (Grace and Bunney, 1983a). The anomalous corded in vivo are under constant bombardment with chloride- rectifier is thought to play a role in maintaining pacemaker firing dependent i.p.s.p.s (Grace and Bunney, 1985), which were not patterns of neurons (Crepe1 and Penit-Soria, 1986) and thus observed to an equivalent extent in the present study, may may contribute to the pacemaker firing pattern observed here. partially account for this difference in input resistance. Indeed, An anomalous rectifier producing a similar, highly prominent Lacey et al. (1988) reported that application of GABA to sub- sag in the membrane potential during hyperpolarization has stantia nigra zona compacta neurons in vitro will reduce their been reported in neocortical pyramidal cells (Schwindt et al., input resistance by 80% with respect to controls. 1988), with this sag in the membrane voltage described as the “hallmark” of a slow anomalous rectifier. This sag apparently Firing pattern is composed of an instantaneous component and a time-depen- A second major difference in the physiology of DA neurons dent component, similar to that described by Yarom and Llinas recorded in vitro versus those reported in vivo is in the firing (1987) in the inferior olive, since both the initial and the steady- pattern observed. It is estimated that about 50-70% of the DA state current-voltage relationships exhibit increasing deviations neurons in the substantia nigra of anesthetized rats are spon- from linearity with larger degrees of hyperpolarization. The taneously active (Grace and Bunney, 1986; Grace, 1987). The anomalous rectification was blocked in DA neurons in vivo by spontaneously active DA neurons recorded in vivo fired in 2 intracellular injection of TEA (Grace and Bunney, 1984a). The distinguishable patterns: (1) a single spiking pattern character- function of the anomalous rectifier in DA cell firing is unclear, ized by individual spikes or doublets occurring with irregular but one possibility is that the action of this conductance may interspike intervals and (2) burst firing, which consists of trains be dependent on the membrane potential. Thus, it should func- of 3-8 spikes of decreasing amplitude and increasing duration tion to decrease excitability in very hyperpolarized DA neurons occurring with comparatively short interspike intervals (i.e., 50- by increasing membrane conductance, while also serving to 150 msec; average: 73 + 13 msec) but with prolonged periods maintain the pacemaker-like activity in spontaneously firing DA of postburst inhibition of activity (approximately 200-450 msec; neurons by keeping the membrane potential positive to the Grace and Bunney, 1984a, b). In contrast, identified DA neurons potassium equilibrium potential (Crepe1and Penit-Soria, 1986; recorded in vitro fired exclusively in a very regular, pacemaker- Schwindt et al., 1988) and possibly closer to the threshold for like pattern, somewhat similar to that found in hypothalamic activating the slow depolarization. histaminergic neurons recorded in vitro (Haas and Reiner, 1988). Furthermore, depolarization of the DA cell increased the fre- Delayed repolarization quency of spike discharge without altering this pacemaker firing In addition to the anomalous rectifier, hyperpolarizing pulses pattern. DA neuron depolarization in vivo typically triggered a revealed the presence of a conductance change that caused a burst firing pattern (Grace and Bunney, 1984b). In our hands, delayed repolarization of the membrane potential. Although the burst firing could not be induced in identified DA neurons in precise nature ofthis event cannot be demonstrated conclusively vitro by any of the wide range of manipulations of membrane without voltage clamping, one interpretation of the results sug- potential or time course of current injections examined in this geststhat this delayed repolarization may be caused by an A-type study. It is unclear why DA neurons in vitro should display such potassium current (IA). This delayed repolarization exhibits markedly different patterns of discharge when compared with properties that are similar to those produced by the well-defined those recorded in vivo. One possibility is the lack of GABAergic I, in other preparations with respect to (1) its activation during bombardment (Grace and Bunney, 1985) and the consequent membrane repolarization from hyperpolarized membrane po- high input resistance found in the in vitro preparation, which tentials, (2) its lack of activation at membrane potentials more could serve to lock the discharge pattern into a very regular negative than - 60 mV, and (3) its insensitivity to TEA (10 mM; structure controlled primarily by the slow depolarization and Connor and Stevens, 1971; Neher, 1971; Thompson, 1977; the afterhyperpolarization. This type of very regular discharge Gustafsson et al., 1982). On the other hand, the insensitivity of pattern is rarely observed in DA neurons recorded in vivo, al- this event to blockade by 4-AP stands in contrast to the reported 3478 Grace and Onn - Identification and Characterization of DA Neurons in vitro pharmacological selectivity of 4-AP for I, channel blockade reflect changes in the threshold of the IS spike. This variability (Connor and Stevens, 197 1; Gustafsson et al., 1982). However, may be due to the comparatively distal location of the IS region a 4-AP-insensitive delayed repolarization with properties oth- with respect to the DA cell soma (Juraska et al., 1977; Preston erwise similar to the I, has been reported in at least one other et al., 198 1; Grace and Bunney, 1983a, b; Tepper et al., 1987). vertebrate neuronal type (hypothalamic histamine-containing This is consistent with previous in vivo studies in which an neurons; Haas and Reiner, 1988). This distinction may be par- electrotonically distal location for the IS spike was inferred by ticularly relevant for DA neurons, since 4-AP appears to block comparing intracellular and extracellular action potentials (Grace a functionally different membrane event that is involved in and Bunney, 1983b). The absence of a prominent afterhyper- regulating DA cell excitability and spike threshold (Grace and polarization following the IS spike combined with the sensitivity Onn, 1988a). of this spike to TTX application is consistent with this being a As reviewed, the putative I, should be triggered following sodium-mediated spike. hyperpolarizing events under conditions that elicit the rebound Blockade of the IS spike and slow depolarization with TTX depolarization. Indeed, coactivation of a putative I, may be one reveal 2 calcium-dependent depolarizations that are involved reason that larger hyperpolarizing current pulses do not trigger in DA neuron spike generation: the low threshold depolarization rebound bursts of spikes in DA neurons, since the proposed and the HTS. The low threshold depolarization in DA neurons, conductance increase could shunt the rebound depolarizing although similar in many respects (e.g., time course, de-inac- events. This stands in contrast to the rebound low threshold tivation with hyperpolarization, sensitivity to cobalt) to the LTS spike (LTS) and burst firing produced at the offset of large mem- described in other vertebrate neurons (Llinas and Yarom, 1981 a, brane hyperpolarizations in some other types of neurons (e.g., b, 1986; Jahnsen and Llinas, 1984a, b; Llinas et al., 1984; Grace Llinas and Yarom, 198 la, b; Grace and Llinas, 1984; Jahnsen and Llinas, 1984), nonetheless shows some unique properties and Llinas, 1984a, b; Nakanishi et al., 1987; Nedergaard et al., in this preparation. Thus, although the rebound LTS in many 1988a, b). This interdependence of the I, and the LTS was vertebrate neurons is mediated primarily if not exclusively by suggestedby Bossu et al. (1985) to account for the responses of calcium, the rebound depolarization in DA neurons appears to sensory neurons to membrane hyperpolarization. In that paper, be composed of both a calcium-dependent low threshold de- the authors concluded that the excitability of neurons could be polarization and a sodium-mediated slow depolarization. The more strongly influenced by either the low threshold, transient LTS has been shown in many other vertebrate neurons to be calcium current (Nowycky et al., 1985) or the I,, depending on important in the regulation of cell excitability. Thus, very hy- the relative level of activation of these currents (i.e., which is perpolarized neurons were found to emit proportionately larger more prominent) at the membrane potential examined. Drawing amplitude LTSs than could be produced in less hyperpolarized from this model, the spike afterhyperpolarization may be of neurons (Llinas and Yarom, 198 la, b, Jahnsen and Llinas, 1984a, sufficient amplitude to activate a rebound low threshold de- b; Grace and Llinas, 1984; Wilcox et al., 1988). However, in polarization, slow depolarization, and spike in depolarized DA DA neurons, the opposite appears to be the case; i.e., whereas cells, whereas stronger hyperpolarizing stimuli should actually smaller hyperpolarizations elicit rebound depolarizations in DA diminish rebound excitation by preferential activation of the I,. neurons at resting or at depolarized membrane potentials, re- Indeed, the voltage-dependent properties of the putative I, could bound depolarizations could not be elicited in more hyperpo- explain why the low threshold depolarizations observed here larized DA neurons. Furthermore, larger hyperpolarizations will are not all-or-nothing events, as reported for the LTS in other actually block activation of the rebound depolarization regard- preparations (Llinas and Yarom, 1981a, b), but instead show less of the initial membrane potential. This blockade could be graded amplitudes that vary with the membrane potential. In due to a depolarization-induced activation of the putative I,. this model, it is possible that the graded amplitude of an all- By increasing the membrane conductance during membrane or-nothing LTS could arise from variations in the level of ac- repolarization, activation of the I, could serve to shunt the tivation of the putative I,. currents mediating the rebound depolarization. Thus, the mem- brane potentials at which the putative LTS can be activated in DA neuron spike subcomponents DA cells is more highly constrained than has been found in The slow depolarization drives the membrane of the DA neuron other neurons. As a consequence, although the low threshold to rather depolarized potentials to trigger spike firing. This in- depolarization could support spontaneous spiking in DA neu- ordinately high spike threshold is also a distinctive feature of rons when it is activated following the spike afterhyperpolari- identified DA neurons recorded in vivo (Grace and Bunney, zation, it probably would play less of a role in a tonically hy- 1984a). The action potential exhibits an inflection in the rising perpolarized DA neuron due to activation of the putative I,. phase of the spike waveform due to the delay between activation A second calcium-dependent depolarization observed in DA of the IS spike and triggering of the somatodendritic (SD) spike, neurons is similar to that described as the HTS in other prep- although it is not as prominent as that observed in vivo (Grace arations (Schwartzkroin and Slawski, 1977; Wong et al., 1979; and Bunney, 1983b). A shortened IS-SD delay is indicative of Llinas and Yarom, 1981a, b; Grace and Llinb, 1984; Kita et a higher safety factor for somatodendritic invasion of the IS al., 1986; Ort et al., 1988). In contrast to the low threshold spike (Renshaw, 1942; Brock et al., 1953) which in this case depolarization, depolarization of TTX-treated DA neurons will may be attributed to the higher membrane resistance of DA not activate HTSs. The HTS can, however, be triggered by neurons in vitro that should facilitate current spread into the comparatively small depolarizations after pretreatment of DA dendrites. The IS spike-generating zone is the lowest threshold neurons with the potassium blocker TEA. The HTS is none- spiking region of vertebrate neurons (Coombs et al., 1957). As theless apparently distinct from the low threshold depolariza- the lowest threshold component of the action potential, the tion, since the low threshold depolarization can still be elicited variability of DA cell spike thresholds at different membrane after TEA using smaller depolarizing pulses. The ability of TEA potentials (Grace and Bunney, 1983b; Grace, 1987) is likely to to reveal the HTS could be accounted for by several mecha- The Journal of Neuroscience, October 1999, 9(10) 3479 nisms; for example, it could block a local resting potassium DA neurons, thereby resetting the membrane potential and de- conductance or delayed rectifier, either of which could conceiv- laying the initiation of a subsequent slow depolarization and ably prevent activation of a nearby spiking zone. However, low threshold depolarization. Blockade of the HTS by cobalt or given (1) the high baseline input resistance of the DA neurons, triggering of the IS spike by antidromic activation allows the (2) the lack of significant alterations in input resistance or cur- isolated IS spike to fire at much higher frequencies than could rent-induced depolarization of the soma by TEA (Fig. 8, B,- normally be attained by a DA neuron firing full amplitude IS- B,), (3) the large degree to which the soma can be depolarized SD action potentials due, in part, to the absence of this con- by current injection in the absence of TEA (Figs. 70, 8B,), and comitant long-duration afterhyperpolarization following these (4) the extent to which TEA lowers spike threshold, the evidence sodium-mediated IS spikes (e.g., Fig. 7c). appears to favor an action of TEA on HTS spike threshold similar to that proposed by Llinas et al. (1984), i.e., that the Spike generation in DA neurons blockade of membrane potassium channels by TEA facilitates Although speculative in nature, the results provide sufficient intraneuronal electrotonic current spread into more distal spike- evidence to construct a preliminary model of spontaneous action generating zones. TEA administration would thereby cause spike potential generation in DA neurons. In this model, the spon- generating zones located in distal dendrites to become compar- taneous slow depolarization, originating near the soma, slowly atively more depolarized in response to a given amount of so- depolarizes the DA neuron membrane until the IS region reaches matic depolarization, causing lowering of the apparent threshold spike threshold. The IS spike then spreads across the soma and of the HTS spike as measured at the soma. Indeed, when one proximal dendrites (already depolarized by the slow depolar- considers that the HTS cannot be activated by large depolari- ization or low threshold depolarization) to trigger the more zations of the somata of DA neurons despite their high input distally located dendritic HTS generation sites. The calcium- resistance, the requirement for TEA to elicit HTSs could indi- mediated HTSs then trigger a calcium-activated afterhyperpo- cate that the HTS spike-generating region is located at a signif- larization that causeswidespread hyperpolarization of the soma icant distance from the soma. One possible location could be and the expansive dendritic tree. The rebound from this after- the distal fork-like dendritic tufts described previously. hyperpolarization could then trigger a rebound low threshold In contrast to its effects on HTS firing, TEA administration depolarization, which would in turn activate the slow depolar- only produces an increase in the amplitude of the low threshold ization and reinitiate the spike generation sequence. In the ab- depolarization without changing its threshold for activation. sence of afferent modulatory activity, this sequence of events, Although not conclusive, this observation, plus the observation coupled with their modulation by the prominent anomalous that it can be elicited by small amplitude depolarizations of the rectifier, could underlie the pacemaker-like spontaneous activity soma, suggeststhat the low threshold depolarization is generated pattern recorded in DA neurons in vitro. Indeed, potentiation at an electrotonically proximal site with respect to the soma and of the effects of these depolarizations by the comparatively high is distinct from the HTS. This may also be true for the slow input resistance could serve to lock the DA neuron into this depolarization, since (1) it can also be activated by brief de- nonvarying pacemaker-like firing pattern. However, the factors polarizations of the soma, and (2) its rate of activation is strongly responsible for modulating this pacemaker pattern into the more dependent on the membrane potential of the soma. If, on the typical irregular and burst firing patterns recorded from DA other hand, these processes were mediated at more electroton- neurons in vivo are unknown at present. ically distal sites, one may predict that longer potential changes or larger currents would be required for their activation. Summary The HTSs are likely to comprise the majority of the SD spike, Despite the differences in the electrophysiological properties of since (1) both the SD spike and the HTS can be blocked by DA neurons recorded in vivo versus those observed in vitro, DA cobalt administration, and (2) TEA was found to increase the neurons can nonetheless be identified in each preparation by duration of the SD and the HTS without affecting the IS spike certain characteristic electrophysiological properties, such as the duration (unpublished observations). During normal spike ac- presence of the slow depolarization, a slow firing rate, a long tivity, the IS spike must be of sufficient amplitude and optimally spike duration, and a high threshold for action potential dis- located to trigger the HTSs. Furthermore, the DA neuron itself charge. The slow depolarization appears to be responsible for must be adequately depolarized to prevent shunting of the IS maintaining spontaneous spike discharge in DA neurons even spike-induced depolarizing current away from the dendrites. in the absence of long-loop afferents and may be a factor in Depolarization of the DA cell soma mediated by the slow de- allowing transplanted DA neurons to retain their intrinsic spike- polarization and the low threshold depolarization would thereby generating activity. facilitate depolarization of the dendrites and triggering of the The in vitro preparation enhances one’s ability to identify the HTS. Thus, antidromic activation (Grace and Bunney, 1983b) endogenous membrane properties of a neuron versus those aris- and the rebound from large hyperpolarizations (Fig. 7E) can ing from the interaction of the neuron with the brain system in trigger the IS spike without activating the SD spike, presumably situ. Thus, although burst firing can be triggered in DA neurons because the soma and proximal dendrites are insufficiently de- in vivo by depolarization (Grace and Bunney, 1984b), recordings polarized (due to the absence of either a preceding spontaneous in the in vitro preparation suggest that a factor that “enables” slow depolarization or rebound low threshold depolarization) the burst firing mode has been removed. If so, this preparation to facilitate the spread of the IS spike into the dendrites. Firing would facilitate inquiries into the mechanism of DA neuron of the calcium-mediated HTS would give rise to the spike af- burst firing. Furthermore, the presence of DA neurons that do terhyperpolarization, which is mediated by a calcium-activated not exhibit spontaneous activity in the in vitro preparation pro- potassium conductance in DA neurons (Grace and Bunney, vides evidence that the state of DA neuron activity may be 1984a). In this way, triggering the HTS would cause a rapid and regulated by factors other than the striatonigral pathway (Grace simultaneous repolarization of the extensive dendritic tree of and Bunney, 1986; Grace, 1987). Comparisons of cellular prop- 3480 Grace and Onn - Identification and Characterization of DA Neurons in vitro

erties across preparations may thereby provide direction to ef- Gerfen, C. R., K. G. Baimbridge, and J. Thilbault (1987b) The neo- forts focused at understanding and eventually modifying the striatal mosaic: III. Biochemical and developmental dissociation of patch-matrix mesostriatal systems. J. Neurosci. 7: 3935-3944. activity of DA neurons in the behaving organism. Grace, A. A. (1987) The regulation of dopamine neuron activity as determined by in vivo and in vitro intracellular recordings. In The References Neurophysiology of Dopamine Systems. L. A. Chiodo and A. S. Free- Adrian, R. H. (1969) Rectification in muscle membrane. Prog. Bio- man, eds., pp. l-67, Lake Shore Publications, Detroit. phys. Mol. Biol. 19: 340-369. Grace, A. A. (1988) In vivo and in vitro intracellular recordings from Bargas, J., E. Galarraga, H. T. Chang, and S. T. Kitai (1988) Electro- dopamine neurons. Ann. NY Acad. Sci. 537: 51-76. physiological and double-labeling immunohistochemical analyses of Grace, A. A., and B. S. Bunney (1980) Nigral dopamine neurons: neurons in the substantia nigra zona compacta of the rat. Sot. Neu- Intracellular recording and identification using L-DOPA injection rosci. Abstr. 14: 1025. combined with fluorescence histochemistry. Science 210: 654-656. Bjorklund, A., and 0. Lindvall (1975) Dopamine in dendrites of sub- Grace, A. A., and B. S. Bunney (1983a) Intracellular and extracellular stantia nigra neurons: Suggestions for a role in dendritic terminals. electrophysiology of nigral dopaminergic neurons. I. Identification and Brain Res. 83: 531-537. characterization. Neuroscience 10: 30 l-3 15. Bossu, J.-L., J.-L. DuPont, and A. Feltz (1985) I, current compared Grace, A. A., and B. S. Bunney (1983b) Intracellular and extracellular to low threshold calcium current in cranial sensory neurons. Neurosci. electrophysiology of nigral dopaminergic neurons. II. Action potential Lett. 62: 249-254. generating mechansims and morphological correlates. Neuroscience IO: Brock, L. G., J. S. Coombs, and J. C. Eccles (1953) Intracellular re- 317-331. cording from antidromically activated motbneurones. J. Physiol. Grace, A. A., and B. S. Bunney (1984a) The control of firing pattern (Lond.) 122: 42946 1. in nigral dopamine neurons: Single spike firing. J. Neurosci. 4: 2866- Brodie, M. S., and T. V. Dunwiddie (1987) Cholecystokinin poten- 2876. tiates dopamine inhibition of mesencephalic dopamine neurons in Grace, A. A., and B. S. Bunney (1984b) The control of firing pattern vitro. Brain Res. 425: 106-l 13. in niaral donamine neurons: Burst firma. J. Neurosci. 4: 2877-2890. Bunney, B. S., and G. K. Aghajanian (1976) D-Amphetamine-induced Grace, A. A., and B. S. Bunney (1985) Opposing effects of striatonigral inhibition of central dopaminergic neurons: Mediation by a striato- feedback pathways on midbrain dopamine cell activity. Brain Res. nigral feedback pathway. Science 192: 391-393. 333: 27 l-284. Bunney, B. S., and A. A. Grace (1978) Acute and chronic haloperidol Grace, A. A., and B. S. Bunney (1986) Induction of depolarization treatment: Comparison of effects on nigral dopaminergic cell activity. block in midbrain dopamine neurons by repeated administration of Life Sci. 23: 1715-1728. haloperidol: Analysis using in vivo intracellular recording. J. Phar- Bunney, B. S., J. R. Walters, R. H. Roth, and G. K. Aghajanian (1973) macol. Exp. Ther. 238: 1092-l 100. Dopaminergic neurons: Effect of antipsychotic drugs and ampheta- Grace, A. A., and R. Llinas (1984) Electrophysiology and morphology mine on single cell activity. J. Pharmacol. Exp. Ther. 185: 560-571. of four classes of prefrontal neocortical neurons studied in vitro. Sot. Cheramy, A., V. Leviel, and J. Glowinski (198 1) Dendritic release of Neurosci. Abstr. IO: 739. dopamine in the substantia nigra. Nature 289: 537-542. Grace, A. A., and R. Llinas (1985) Dehydration-induced morpholog- Cheramy, A., R. Romo, and J. Glowinski (1985) Dopamine release ical artifacts in intracellularly stained neurons: Circumvention using and neuronal activity in the two nigrostriatal systems. In In Viva rapid DMSO clearina. Neuroscience 16: 461-475. Perfusion and Release of Neuroactive Substances, A. Bayon and R. Grace, A. A., and S.-P. &n (1988a) Regulation of sodium and calcium Drucker-Cohn, eds., pp. 161-175, Academic, New York. spikes in immunocytochemically identified rat dopamine neurons. Chiodo, L. A. (1988) Dopamine-containing neurons in the mamma- Sot. Neurosci. Abstr. 14: 75. lian central nervous system: Electrophysiology-. and pharmacology.-. Grace, A. A., and S.-P. Onn (1988b) Biphasic regulation of activity Neurosci. Biobehav. Rev. 12: 49-9 1. in immunochemically identified rat midbrain dopamine neurons re- Chiodo. L. A.. and B. S. Bunnev (1983) Tvoical and atvoical neuro- corded in vitro. J. Neurosci. Methods 24: 190. leptics: Differential effects of chronic admmistration on-the activity Greenfield, S. A., J. J. B. Jack, A. T. J. Last, and M. French (1988) of A9 and A10 midbrain dopaminergic neurons. J. Neurosci. 3: 1607- An electrophysiological action of acetylcholinesterase independent of 1619. its catalvtic site. Exp. Brain Res. 70: 441-444. Chiodo, L. A., and G. Kapatos (1987) Mesencephalic dopamine-con- Groves, P: M., and J.C. Linder (1983) Dendro-dendritic synapses in taining neurons in culture: Morphological and electrophysiological substantia nigra: Descriptions based on analysis of serial sections. characterization. In The Neurophysiology of Dopamine Systems, L. Exp. Brain Res. 49: 209-217. A. Chiodo and A. S. Freeman, eds., pp. 67-91, Lake Shore Publi- Groves, P. M., C. J. Wilson, S. J. Young, and G. V. Rebec (1975) Self- cations, Detroit. inhibition by dopaminergic neurons. Science 190: 522-529. Connor, J. A., and C. F. Stevens (1971) Voltage clamp studies of a Gulley, R. L., and R. L. Wood (197 1) The fine structure of the neurons transient outward membrane current in gastropod neural somata. J. in the rat substantia nigra. Tissue Cell 3: 675-690. Physiol. (Lond.) 213: 21-30. Gustafsson, B., M. Galvan, P. Grafe, and W. Wigstrom (1982) A Coombs, J. S., D. R. Curtis, and J. C. Eccles (1957) The interpretation transient outward current in a mammaliam central neurone blocked of spike potentials of motoneurones. J. Physiol. (Lond.) 139: 198- by 4-aminopyridine. Nature 299: 252-254. 231. Guyenet, P. G., and G. K. Aghajanian (1978) Antidromic identifi- Crepel, R., and J. Penit-Soria (1986) Inward rectification and low cation of dopaminergic and other output neurons of the rat substantia threshold calcium conductance in rat cerebellar Purkinje cells. J. Phys- nigra. Brain Res. 150: 69-84. iol. (Lond.) 372: l-23. Guyenet, P. G., and J. K. Crane (198 1) Non-dopaminergic nigrostri- Deniau, J. M., C. Hammond, A. Riszk, and J. Feger (1978) Electro- atal pathway. Brain Res. 213: 291-305. physiological properties of identified output neurons of the substantia Haas, H. L., and P. B. Reiner (1988) Membrane properties ofhistamin- nigra pars compacta and pars reticulata: Evidence for the existence ergic tuberomammillary neurones of the rat hypothalamus in vitro. of branched projections. Exp. Brain Res. 32: 409-422. J. Physiol. (Lond.) 339: 633-646. Fallon, J. H., J. N. Riley, and R. Y. Moore (1978) Substantia nigra Hokfelt, T., 0. Johansson, K. Fuxe, M. Goldstein, and D. Park (1976) dopamine neurons: Separate populations project to neostriatum and Immunohistochemical studies on the localization and distribution of allocortex. Neurosci. Lett. 7: 157-l 62. monoamine neuron systems in the rat brain. I. Tyrosine hydroxylase Fuxe, K., M. Goldstein, T. HSkfelt, and T. H. Joh (1971) Cellular in the mes- and diencephalon. Med. Biol. 54: 427-453. localization of dopamine-@-hydroxylase and phenylethylamine-N- Homvkiewicz. 0. (1963) Die tonische lokalisation und das verhalten methyl transferase as revealed by immunohistochemistry. Prog. Brain von noradrenalin und dopamin-(3-hydroxytyramin) in der substantia Res. 34: 127-138. nigra des normalin und Parkinsonkranken. Menschen. Wien. Klin. Geffen, L. B., T. M. Jessel, A. C. Cuello, and L. L. Iversen (1976) Wochenschr. 75: 309-355. Release of dopamine from dendrites in rat substantia nigra. Nature Homykiewicz, 0. (1966) Dopamine (3-hydroxytyramine) and brain 260: 258-260. function. Pharmacol. Rev. 18: 925-964. Gerfen, C. R., M. Herkenham, and J. Thilbault (1987a) The neostriatal Jahnsen, H., and R. Llinas (1984a) Electrophysiological properties of mosaic: II. Patch- and matrix-directed mesostriatal dopaminergic and guinea-pig thalamic neurones: An in vitro study. J. Physiol. (Lond.) non-dopaminergic systems. J. Neurosci. 7: 39 15-3934. 349: 205-226. The Journal of Neuroscience, October 1989, 9(10) 3481

Jahnsen, H., and R. Llinls (1984b) Ionic basis for the electrorespon- and electrophysiology of ventral mesencephalon nerve cell cultures. siveness and oscillatory properties of guinea pig thalamic neurones Dev. Brain Res. 39: 205-2 15. in vitro. J. Physiol. (Lond.) 349: 227-247. Paxinos, G., and C. Watson (1982) The Rat Brain in Stereotaric Jiminez-Castellanos, J., and A. M. Graybiel (1987) Subdivisions of Coordinates, Academic, New York. the dopamine-containing A8-A9-A10 complex identified by their dif- Pinnock, R. D. (1984) Hyperpolarizing action of bachlofen on neurons ferential mesostriatal innervation of striosomes and extrastriosomal in the rat substantia nigra slice. Brain Res. 332: 337-340. matrix. Neuroscience 23: 223-242. Pinnock, R. D. (1985) Neurotensin depolarizes substantia nigra do- Jones. E. G. (1975) Varieties and distribution of non-ovramidal cells pamine neurons. Brain Res. 338: 15 l-l 54. in the somatic sensory cortex ofthe squirrel monkey. J.-Camp. Neurol. Preston, R. J., R. A. McCrea, H. T. Chang, and S. T. Kitai (198 1) 160: 205-268. Anatomy and physiology of substantia nigra and retrorubral neurons Juraska, J. M., C. J. Wilson, and P. M. Groves (1977) The substantia studied by extra- and intracellular recording and by horseradish per- nigra of the rat: A Golgi study. J. Comp. Neurol. 172: X5-600. oxidase labeling. Neuroscience 6: 33 l-344. Katz, B. (1949) Les constantes electriques de la membrane du muscle. Renshaw, B. (1942) Effects of presynaptic volleys on spread of im- Arch. Sci. Physiol. 3: 285-300. pulses over the soma of the motoneuron. J. Neurophysiol. 5: 235- Kita, T., H. Kita, and S. T. Kitai (1986) Electrical membrane prop- 243. erties of rat substantia nigra compacta neurons in an in vitro slice Rinvik, E., and I. Grofova (1970) Observations on the fine structure preparation. Brain Res. 372: 21-30. of the substantia nigra in the cat. Exp. Brain Res. 11: 229-248. Kitahama, K., L. Denoroy, M. Goldstein, M. Jouvet, and J. Pearson Romo. A.. A. Chtramv, G. Godeheu. and J. Glowinski (1986) In vivo (1988) Immunohistochemistry of tyrosine hydroxylase and phenyl- presynaptic control of dopamine release in the cat caudate nucleus- ethanolamine N-methyltransferase in the human brain stem: De- I. Opposite changes in neuronal activity and release evoked from scription of adrenergic perikarya and characterization of longitudinal thalamic motor nuclei. Neuroscience 19: 1067-1079. catecholaminergic pathways. Neuroscience 25: 97-l 11. Ruffieux, A., and W. Schultz (1980) Dopaminergic activation of re- Korf, J., M. Zieleman, and B. H. C. Westerink (1976) Dopamine ticulata neurones in the substantia nigra. Nature-285: 240-24 1. release in the substantia niara. Nature 260: 257-258. Schwartzkroin. P. A., and M. Slawski (1977) Probable calcium snikes Lacey, M. G., N. B. Mercuri,>nd R. A. North (1987) Dopamine acts in hippocampal neurons. Brain Res. ‘135: i 57-l 6 1. on-D2 receptors to increase potassium conductance in neurons of the Schwindt, P. C., W. J. Spain, and W. E. Crill (1988) Influence of rat substantia niara zona comoacta. J. Phvsiol. (Land.) 392: 397-4 16. anomalous rectifier activation on afterhyperpolarizations of neurons Lacey, M. G., N. By Mercuri, and R. A. North (19‘88) On the potassium from cat sensorimotor cortex in vitro. J. Neurophysiol. 59: 468-48 1. conductance increase activated by GABA, and dopamine D, recep- Schwyn, R. C., and C. A. Fox (1974) The primate substantia nigra: tors in rat substantia nigra neurones. J. Physiol. (Lond.) 401: 437- A Golgi and electron microscopic study. J. Hirnforsch. 16: 95-126. 453. Shepard, P. D., and B. S. Bunney (1988) Effects of apamin on the Lindvall, O., and A. Bjijrklund (1974) The glyoxylic acid fluorescence discharge properties of putative dopamine-containing neurons in vi- histochemical method: A detailed account of the methodology for the tro. Brain Res. 463: 380-384. visualization of central catecholamine neurons. Histochemistry 39: Silva, N. L., and B. S. Bunney (1988) Intracellular studies of dopamine 97-127. neurons in vitro: Pacemakers modulated by dopamine. Eur. J. Phar- Lindvall, O., and A. BjGrklund (1978) Anatomy of the dopaminergic macol. 149: 307-3 15. neuron systems in the rat brain. Adv. Biochem. Psychopharmacol. Snyder, S. H. (1973) Amphetamine psychosis: A model schizophrenia 19: 139-231. mediated by catecholamines. Am. J. Psychiatry 130: 61-67. Llinas, R., and M. Sugimori (1980a) Electrophysiological properties Sternberger, L. A., P. H. Hardy, J. J. Cuculis, and H. G. Meyer (1970) of in vitro Purkinje cell somata in mammalian cerebellar slices. J. The unlabeled antibody enzyme method of immunocytochemistry. Physiol. (Lond.) 305: 17 l-l 95. Preparation and properties of soluble antigen-antibody complex Llinls, R., and M. Sugimori (1980b) Electrophysiological properties (horseradish peroxidase-antihorseradish peroxidase) and- its use in of in vitro Purkinje cell dendrites in mammalian cerebellar slices. J. identification of soirochetes. J. Histochem. Cvtochem. 18: 3 15-333. Physiol. (Lond.) 305: 197-2 13. Stewart, W. W. (1978) Functional connections between cells as re- Llinls, R., and Y. Yarom (1981a) Electrophysiology of mammalian vealed by dye-coupling with a highly fluorescent naphthalimide tracer. inferior olivary neurones in vitro. Different types of voltage-dependent Cell 14: 741-759. ionic conductances. J. Physiol. (Lond.) 315: 549-567. Swanson, L. W., and B. K. Hartman (1975) The central adrenergic Llinls, R., and Y. Yarom (198 1b) Properties and distribution of ionic system. An immunofluorescence study of the location of cell bodies conductances generating electroresponsiveness of mammalian infe- and their efferent connections in the rat using dopamine-fl-hydrox- rior olivary neurones in vitro. J. Physiol. (Lond.) 315: 569-584. ylase as a marker. J. Comp. Neurol. 163: 467-506. Llinas, R., and Y. Yarom (1986) Oscillatory properties of guinea-pig Tepper, J. M., S. F. Sawyer, and P. M. Groves (1987) Electrophys- inferior olivaty neurones and their pharmacological modulation: An iologically identified nigral dopaminergic neurons intracellularly la- in vitro study. J. Physiol. (Lond.) 376: 163-182. beled with HRP: Light-microscopic analysis. J. Neurosci. 7: 2794- Llinls, R., S. A. Greenfield, and H. Jahnsen (1984) Electrophysiology 2806. of pars compacta cells in the in vitro substantia nigra: A possible Thierry, A. M., J. M. Deniau, and J. Feger (1979) Effects of stimulation mechanism for dendritic release. Brain Res. 294: 127-132. of the frontal cortex on identified output VMT cells in the rat. Neu- Matthysse, S. (1973) Antipsychotic drug actions: A clue to the neu- rosci. Lett. 15: 103-107. ropathology of schizophrenia? Fed. Proc. 32: 200-205. Thompson, S. H. (1977) Three pharmacologically distinct potassium Nakanishi, H., H. Kita, and S. T. Kitai (1987) Intracellular study of channels in molluscan neurones. J. Physiol. (Lond.) 265: 465-488. rat substantia nigra pars reticulata neurons in an in vitro slice prep- Waszczak, B. L., and J. R. Walters (1983) Dopamine modulation of aration: Electrical membrane properties and response characteristics the effects of gamma aminobutyric acid on substantia nigra pars re- to subthalamic stimulation. Brain Res. 437: 45-55. ticulata neurons. Science 220: 2 18-22 1. Nedergaard, S., J. P. Bolam, and S. A. Greenfield (1988a) Facilitation White, F. J., and R. Y. Wang (1983) Comparison of the effects of of a dendritic calcium conductance by 5-hydroxytryptamine in the chronic haloperidol treatment on A9 and A10 dopamine neurons in substantia nigra. Nature 333: 174-177. the rat. Life Sci. 32: 983-993. Nedergaard, S., C. Hopkins, and S. A. Greenfield (1988b) Do nigro- Wilcox, K., M. J. Gutnick, and G. R. Christoph (1988) Electrophys- striatal neurones possessa discrete dendritic modulatory mechanism? iological properties of neurons in the lateral habenula nucleus: An in Electrophysiological evidence from the actions of amphetamine in vitro study. J. Neurophysiol. 59: 212-225. brain slices. Exp. Brain Res. 69: 444-448. Wong, R. K. S., D. A. Prince, and A. I. Busbaum (1979) Intradendritic Neher, E. (197 1) Two fast transient current components during voltage recordings from hippocampal neurons. Proc. Natl. Acad. Sci. USA clamp on snail neurons. J. Gen. Physiol. 58: 36-53. 76: 986-990. Nowycky, M. C., A. P. Fox, and R. W. Tsien (1985) Three types of Yarom, Y., and R. Llinas (1987) Long-term modiEability of anom- neuronal calcium channel with different calcium agonist sensitivity. alous and delayed rectification in guinea pig inferior olivary neurons. Nature 316: 440-443. J. Neurosci. 7: 1166-l 177. Ort, C. A., K. J. Futamachi, and J. H. Peacock (1988) Morphology